EVALUATING THE MANDATORY ACTIVATION OF OLDER UNEMPLOYED
Author: Joost Bollens
1.
Abstract
In the course of May 2009 the so-called systematic approach by the VDAB (The Flemish Service for Employment and Vocational Training or the Flemish Public Employment Service) was extended to the group of the newly unemployed between the ages of 50 and 52 years. This approach means that these unemployed are required to seek assistance to find a new job after three months of unemployment. In so doing they can rely on various instruments, such as participation in a 50+ club and other types of support for finding a job, education programs, etc. They are also assisted by consultants who only work with jobseekers aged 50 or older and who also can take into account the specific issues of this target group. This study examines whether this measure has had an impact on the career opportunities of this target group. In isolating this impact we used the fact that the introduction of the measure is based on an age criterion (50-52 years) and on a calendar criterion (from May 2009 onwards). The conclusion is that the measure has increased the transition to employment in this target group by 3 to 4 percentage points, which, in view of the relatively low chances of outflow to employment of jobseekers aged 50 or older, is considerable. Moreover there are also indications that not all the unemployed in this target group have effectively made use of this measure in spite of the mandatory nature of this measure. This suggests that the impact for those who did take part is potentially deemed even higher than was estimated.
Key words: Active labour market policy; older unemployed; difference in differences; propensity score matching;
2.
Study objectives
In view of the demographic trends it is very important that a sufficiently high employment rate is maintained in the category of the over-50s. This is more particularly an issue in the Flemish labour market, which is characterised by a very low employment rate in this age group. This can be remediated first and foremost by ensuring that the necessary regulations and conditions are in place to encourage older employees to work longer and to encourage employers to keep older employees on board longer. One must then subsequently ensure that the over 50s who do indeed become unemployed can rely on adapted tools to help them find a job again. These include the tools of the active labour market policy, which can be adapted where necessary to this age group’s specific issues and needs. In spite of the fact that there is a lot of information about the effectiveness of various active labour measures in general, there is not much information about the impact of measures that specifically target older jobseekers. In addition, in those cases where we can say something about the relation between effectiveness and age, based on literature, this usually relates to older jobseekers who elected to make use of the measure themselves, and which in all probability form a selective, non-representative group. The measure studied here is quite important in this respect, not only because it involves tools that have been adapted in function of the age group (to a certain extent, among others by using specific consultants) but because anyone who falls under these conditions is supposed to take advantage of the measure. The aim of the study was to verify whether the introduction of this new policy has had an impact on the job opportunities of this target group.
3.
Methods and data
We explain here in five steps how the measure’s impact was determined. (1) Firstly, the group of all the jobseekers in the age group (50≤ age < 53) who were eligible for the measure since the start of the extended systematic approach was selected. This is called the target group hereinafter. The labour market trajectory of these people was then monitored on a monthly basis, from the instant that they exceeded the threemonth period of unemployment (from this time onwards they were eligible for the measure) for a seven-month period (meaning until the tenth month after they became unemployed). Every month we verified whether or not they were working. It is then possible to calculate the proportion of this group that is employed for a seven-month period. (2) Subsequently the employment results of the target group were compared with the employment results of a group of jobseekers in the 50-52 age group, who became unemployed in the same months as the target group, albeit in 2008
1
instead of in 2009. There was no 50+ systematic approach for these jobseekers who became unemployed in 2008 so that this systematic approach constitutes one of the explanations for the potential differences between their performance and that of the target group. (3) But probably several other aspects changed between 2008 and 2009 in addition to the introduction of this new policy, or even: Difference between 2009 and 2008 = (effect of the systematic approach) + (effect of other factors) In a third stage the impact of these other factors was estimated. Once this impact is known the impact of the systematic approach can be isolated in the comparison. In order to assess how the job performance of the recently unemployed subjects in the 50-52 age group would have evolved between 2008 and 2009 if the systematic approach had not been introduced, we chose a nearby group, made up of subjects from the age groups near 50-52 years, who recently became unemployed and who did not benefit from the systematic approach in the period studied. The subjects who became unemployed in 2008 and 2009 in the (53 ≤ age < 55) age group comply with this description. (4) It is rather probable that the composition of the inflow of subjects in the 50-52 age group who were unemployed in 2009 was different compared with 2008. In order to take this into account the composition of all the groups was equalised by using matching techniques, prior to steps (1) through (3). (5) In step (3) the so-called "common trend" assumption was made: in the absence of the policy intervention both groups (50-52 and 53-55) would have developed according to the same trend between 2008 and 2009. It is impossible to test this assumption but we were able to check whether it was plausible, among others by checking whether the same trend development could be observed in another nearby comparison group as in the comparison group of 53-55 year-olds. To this end we studied jobseekers who became unemployed between the ages of 48 and 50 years, again in 2008 and in 2009. In 2009 these jobseekers also fell under the systematic approach, but this was already the case in 2008, so it is safe to say that there is no change for this group between 2008 and 2009.
4.
Findings
The jobseekers who fall under the systematic approach were subsequently monitored for a seven-month period. Every month we checked whether they were employed or unemployed. There were also data available for sevenmonth periods for the comparison groups (50-52 in 2008, 53-55 in 2008 and 53-55 in 2009). Table 1 shows the estimated impact for these seven month-periods, obtained according to the reasoning explained in section 3, resulting in a difference in difference after matching. The upper part of the table relates to the approach for which the correction for the trend evolution between 2008 and 2009 was made based on the 53-55 yr old comparison groups. These results suggest that the systematic approach had an impact on jobseekers’ chances of finding a job again. This impact is visible, but with some delay. In month 4, one month after the jobseeker becomes eligible for the measure, the impact is still nil. It then takes a few months before the estimated impact is sufficiently high to be considered as statistically significant and different from nil. This is the case from month 7 onwards. The estimated impact for month 9, for example, means that, as a result of the systematic approach, employment rises by 4.5 percentage points. In view of the fact that 24.6% of jobseekers in the group falling under the systematic approach were effectively employed in month 9 this result suggests that in the absence of this measure only (24.6%-4.5%=) 20.1% would be employed. Table 1 Difference-in-differences matching results (treated individuals : n = 3908) Month 4
Month 5
Month 6
Month 7
Month 8
Month 9
Month 10
Second difference : Group 53-55 Effect
-0.001
0.018
0.015
0.028
0.034
0.045
0.041
Bootstrapped Standard Error*
0.01
0.012
0.013
0.014
0.014
0.015
0.016
T-value
-0.14
1.47
1.09
1.96
2.30
2.97
2.54
Second difference : Group 48-50 Effect
0.002
0.010
0.011
0.016
0.022
0.026
0.028
Bootstrapped Standard Error*
0.010
0.012
0.013
0.014
0.015
0.015
0.016
T-value
0.25
0.86
0.80
1.10
1.48
1.68
1.82
* with 400 replications
2
Chart 1 Estimated impact of the systematic approach It is possible to deduce from the table that the estimates deviate from the previous results, based on this alternative estimate strategy. A comparison of both sets in Chart 1 effectively indicates a difference but both sets are characterised by a marked upward trend. In that sense the common trend assumption is not refuted. As far as the size of the estimated effect is concerned it does of course make a difference which of the two sets is taken into consideration. The researchers argue that the comparison with the group of 53-55s is the most relevant comparison, among others because the age of 50 years is an important psychological barrier in the Belgian labour market. The question remains then whether or not the observed effect is substantial. Ultimately this question can only be effectively answered if one takes the cost of the measure into account because a tool’s impact needs to be related to the cost of its implementation. Detailed cost figures are not available however.
Chart 2 The observed proportions of employed subjects and the estimated proportions in absence of the measure In Chart 2 the top line indicates which percentage of the target group was effectively employed in months 4 through. 10. The corresponding estimated policy impact has then been deducted every month in order to arrive at the lower curves. The bottom curve indicates the estimated proportions of employment if the measure had not been introduced (based on a trend correction using the results for the 53-55 age group). From this one can deduce that the estimated impacts, in light of the relatively low chances of older jobseekers to find employment, are certainly substantial. Moreover there are also indications that not all the unemployed in this target group have effectively made use of this measure in spite of its mandatory nature. This suggests that the impact for those who did take part is potentially deemed even higher than was estimated.
3
5.
Conclusions and policy implications
In the course of May 2009 the so-called systematic approach in Flanders was extended to the group of the newly unemployed between the ages of 50 and 52 years. This approach means that these unemployed are required to seek assistance to find a new job after three months of unemployment. In so doing they can rely on various instruments, such as participation in a 50+ club and other types of support for finding a job, education programs, etc. They are also assisted by consultants who only work with jobseekers aged 50 or older and who also can take into account the specific issues of this target group. This study examines whether this measure has had an impact on the career opportunities of this target group. When isolating this impact we used the fact that the introduction of the measure is based on an age criterion (50-52 years) and on a calendar criterion (from May 2009 onwards). The fact is that the measure has raised the transition into employment within this target group by 3 to 4 percentage points, which, in views of the relatively low probability of an outflow into employment among jobseekers aged 50 or older is significant. Moreover there are also indications that not all the unemployed in this target group have effectively made use of this measure in spite of its mandatory nature. This suggests that the impact for those who did take advantage of it is potentially deemed even higher than was estimated. These results are important because they indicate that a policy aimed at the activation of older unemployed can still be successful. The policy recommendations, as a result, are clear: a gradual extension of these approaches to those unemployed who are aged 52 or older may certainly be considered. We do recommend, however, that the impact of the measure be monitored. Moreover, we also recommend a better follow-up of which jobseekers in the target group effectively make use of this measure. In so doing the possibility that a better follow-up of mandatory participation may result in a reduction of the measure’s impact. After all, one can argue that especially those unemployed who have taken part in the scheme are the subjects who think that they stand to benefit the most.
Full reference of the study report Bollens, J. (2011). Evaluating the mandatory activation of older unemployed. WSE Report, Leuven.
4